CHAPTER 5: INTERVIEW STUDY
5.2 Interview Instruments and Data Analysis Procedures
5.2.5 Qualitative Content Analysis
5.2.5.1 The Concept of Qualitative Content Analysis
Content analysis was chosen as an inductive approach to the interview data analysis and model building, with the expectation that clear concepts and propositions will emerge as a result. Content analysis is a popular approach to the analysis of both quantitative and qualitative information and a potent technique for researchers to understand what is there (Adams, et al., 2010). As the name suggests, the purpose of content analysis is to describe the content of the respondents’
comments systematically and to classify the various meanings expressed in the material that has been recorded.
In particular, this research adopted qualitative content analysis with an integrated vision of speech and the specific context, in contrast with the traditional quantitative approach where themes and patterns manifest through merely counting words or extracting objective content. Hsieh & Shannon (2005) define it as a research method “for the subjective interpretation of the content of text data through the systematic classification process of coding and identifying themes or patterns”.
Because qualitative content analysis tends to involve purposively selected text to address the research questions, it is able to avoid the situation where syntactical and semantic information embedded in the text is missing from a typical quantitative approach. Additionally, qualitative content analysis is mainly inductive, and therefore more appropriate for generating theories, which prepares the formulation of a preliminary framework as designated in the research design. Moreover, (Smith, 1975) suggested that qualitative content analysis deals with the antecedent-consequent patterns of form, while the quantitative approach deals with duration and frequency of form. The former captures informants’ descriptions or expressions in detail and highlights the extraction of unique themes that could expand the connotation of the researched subject, rather than the statistical significance of the occurrence of presumed concepts. This feature caters for the particular purpose of this interview study to explain and distil the questionnaire survey findings that were produced mostly with the statistical approach.
5.2.5.2 The Process of Qualitative Content Analysis
To support valid and reliable inferences, qualitative content analysis generally entails eight systematic steps for processing the data: (1) prepare the data; (2) define
the unit of analysis; (3) develop categories and a coding scheme; (4) test the coding scheme on a sample of text; (5) code all the text; (6) assess coding consistency; (7) draw conclusions from the coded data; and (8) report the methods and findings. The eight steps lead to a focus on identifiable themes and patterns and gradually reduce the interview data into areas relating to the purpose of the study. This process can be flexible depending on the particular purpose of the study (Zhang & Wildemuth, 2009). The qualitative content analysis in this research adopts six standardised steps based on the particular nature of this research as follows:
Step1: Prepare the Data
Twenty interviews audios were fully transcribed into approximately 250 pages in Word files before analysis started. The 20 Word files were then imported into the QSR NVivo 9 program for coding in the subsequent steps.
Step 2: Define the Unit of Analysis
Instead of using physical linguistic units like a word, sentence or paragraph, qualitative content analysis often employs individual themes as the unit of analysis.
Instances of themes could include any linguistic unit as long as the analyst is primarily looking for the expression of an idea (Minichiello, 1990). Accordingly, the unit of analysis of this interview study was defined by reference to pre-designed themes such as roles, work process, benefit, risk and collaboration.
Step 3: Develop Coding Schemes
The researcher predetermined five coding categories based on the review of the literature and the results of the questionnaire survey. The first stage of coding was accordingly conducted deductively until new, related themes emerged and the inductive approach was used. The second stage of coding then followed with the aid of the constant comparative method to reaffirm existing categories, integrate related categories and seize emerging categories (Glaser et al., 1968). With the assistance of the QSR NVivo9 program, 10 nodes representing 10 coding categories were created as shown in Table 5.5. Meanwhile, all the coded ideas were assigned with a
“stakeholder” attribute, and eventually distilled as the essential findings under each of the seven key stakeholder groups. For example, all the discussion regarding the potential mutual benefits of engaging in sustainable housing was coded under the
“MB” node and classified under specific stakeholders.
Table 5.5 NVivo Coding Summary
Coding nodes Number of references from coded sources
Number of coding sources
Work process and roles 97 15
Benefit gain 49 13
Benefit loss (risk) 34 12
Current collaboration 34 8
Potential collaboration 12 8
Mutual benefits paradigm 25 10
CAB practices & solutions 100 13
Other issues raised 35 12
Behaviour change 52 13
New paradigm 55 12
Step 4: Test Coding Scheme on a Sample
The coding consistency of the first coded theme, “work process”, was checked.
No major adjustment was made.
Step 5: Code All the Text
All the 20 interviews were carefully examined and the core ideas were excerpted and coded under the 10 themes without any significant new themes emerging. The 10 themes were eventually sorted into four categories in the interview report (presented in detail in Appendix B3).
Step 6: Draw Conclusions from the Coded Data
The last step involves interpreting the themes identified and exploring their properties and dimensions (Bradley, 1993). The outcome of this step is reported in detail in the following sections.
5.2.5.3 The Validity of Qualitative Content Analysis
There are four criteria for evaluating interpretive research work such as qualitative content analysis: credibility, transferability, dependability and conformability (Bradley, 1993; Lincoln & Guba, 1985). Table 5.6 provides a description of each criteria and the techniques this interview study adopted to establish these criteria.
Table 5.6 Criteria and Validity of Qualitative Content Analysis of this Research
Criteria Description Techniques used to establish criteria
Credibility Adequate representation of the constructions of the social world under study.
Triangulate interviewee response
Use precise coding definitions and clear coding procedures
Select high-profile interviewees Transferability The extent to which the
researcher’s working
hypothesis can be applied to another context.
Provide rich datasets and descriptions of how data evolved to knowledge (Appendix B2) Dependability The coherence of the internal
process and the way the researcher accounts for changing conditions in the phenomena.
Adopt six standardised steps to ensure the consistency of the study process
Confirmability The extent to which the characteristics of the data, as posited by the researcher, can be confirmed by others who read or review the research results.
Distinguish data, information, knowledge (findings) and wisdom
(outcome/recommendation) (as discussed in Section 3.4.5)